基于地形随机场模型的遥感图像几何纠正

从地形平稳随机场模型出发,应用空间统计学中的Kriging插值算法,提出在多项式纠正基础上进行误差改正的图像精纠正方法,并进行相应的实验验证。多项式纠正和基于共线方程的几何成像模型纠正是最常用的两类纠正方法,但多项式纠正无法克服由于地形起伏所引起的投影差。提出的方法和流程,不仅能克服多项式模型中的弊端,而且不额外增加控制点选择和计算的负担,可以作为多项式纠正流程中的求精。

Geometric Rectification of Remotely Sensed Imagery Based on Random Field Modeling of Topography

Based on the modeling of the stationary random field property of topography,a novel method for geometric rectification of remotely sensed imagery is discussed. The new method establishes the model of error caused by polynomial rectification as a stationary random field.By resorting to Kriging interpolation,errors are estimated pixel by pixel,which are used for refining the rectification result by common polynomial approach.Two experiments are carried out,both show the effectiveness of the proposed method.Since polynomial rectification requires only simple ancillary data,and the algorithm is quite easy to implement,this method is widely used for rectifying remote sensing image from no matter what kind of sensors the image is acquired.However,errors caused by relief can not be removed completely by polynomial model.Though in extended polynomial model height is added as one argument,it is actually difficult to explain why relief errors can be removed.Method proposed in this paper overcomes the drawback of polynomial model,while no extra works are needed for ground control points selection and calculation,so it should be a refinement tool after polynomial rectification is carried out.